The Last Look…

Following my column to start the year, I am indebted to a friend who shared some high level data with me regarding reject rates – even though it may shoot my theory down in flames (not that that has never happened before).

I pointed out that fill rates on three venues, two of which publish the data, had dropped over the last six months of the year, a period that coincided with new LPs emerging in FX. In fairness, I did not lay all the blame at the feet of these firms, and the new data suggests it is part of a broader trend – albeit one with an explanation of sorts.

I checked in with a couple of other “friendlies” and got the same response, so at a high level, what I am being told is that reject rates on aggregators, are close to double what they were around Q3 2024. I should stress, these rates are still low, 3-4%, but that is a noticeable jump from the 1-2% registered for much of the recent past. These rates are also significantly lower than the 10-25% seen on several of the multi-dealer platforms.

The three sources of information all use aggregation and all have a small number of LPs on there (four-to-eight, not universally supporting all currency pairs), so they are not in the 25-LP game, where I would expect reject rates to be higher. They are also, I believe, generally seen as “good” customers engaged in sustained relationships.

Should we, therefore, be concerned about a rise in reject rates from, say, 1.5% to 3%? Markets are more nervous, and that makes LPs nervous too, and more prone to tighten their last look criteria, but does it justify a doubling in rejects? The circumstances are clearly going to be different, so it is hard to put a firm view on this – the three consumers use one or both of sweep and full amount strategies – but I think while it is understandable, I am less sure it is a wise move on the part of the LPs.

Yes, their reject rates are much lower on the bilateral streams, but a doubling of the reject rate could easily mean a doubling (or more) of the cost of rejects, which is what we really should be concerned about. If the higher reject rates result in discernibly higher costs, are the consumers going to be driven into the arms of the multi-dealers?

This is a prime example of why consumers in FX should use the analytical tools available properly, to determine the real cost in dollars and cents. It’s not as though it can’t be done, XTX-Ray must have been produced seven or eight years ago if my memory serves me well, and other (independent) firms have followed suit, so the data is available, if someone actually wants to use it.

There is, I think, an explanation of sorts in the data that was shared with me – the actual “event” month of November, the US election of course, seems to have had lower reject rates, whereas the more random nature of events in January have seen rejects climb. This tells me that the FX market remains excellent at preparing for “known-unknowns”, but it also offers a lesson.

Surely what happened in November was LPs widened out slightly? They knew there was an uncertain (binary) outcome to the election, and prepared accordingly – as we now live in a more chaotic geo-political environment, surely the answer should be slightly wider prices, rather than reject rates?

The fact is, a great number of liquidity consumers are exiting their own risk acquired from their downstream counterparties, so they will merely pass on wider rates, which is fair. As far as the “look” is concerned, far better to have slightly, but barely-discernible wider spreads, than a doubling of reject rates? I am not sure how many liquidity consumers have become obsessed with the spread, but surely the price is more important in a busy market? Ergo, they can accept slightly wider prices to avoid the huge uncertainty of higher rejects.

LPs have the right to reject given the modern FX market structure, but last look has been a thorn in side of the industry for a long time now – so why are so few actually calculating the real cost of rejects? If they are and believe it a price worth paying; fine, but what bothers me is the number of clients who complain (this does not include those I got data from last week to be clear) without actually using such an important datapoint.

To my mind, a 3% reject rate on a bilateral, aggregated stream, is too high, albeit, to repeat myself, understandable. Equally, I find a 15% reject rate on a multi-dealer venue unacceptable, but, again, understandable.

Liquidity provision is a service, and that means it does not, or should not, come free, but perhaps too many are obsessed with the spread because it “looks good”. Surely it is better to be armed with good data on the cost of rejects and use that to establish a firmer, more consistent relationship – one in which you know you will occasionally be rejected but can be confident it will be a rare event?

To my mind, a 3% reject rate on a bilateral, aggregated stream, is too high, albeit, to repeat myself, understandable. Equally, I find a 15% reject rate on a multi-dealer venue unacceptable, but, again, understandable. I cannot escape the feeling that the LPs (and I would argue that they should understand that sometimes a customer is just going to get lucky with their timing, and that should not be punished with a reject) are overusing last look and missing an opportunity to stand out from the field – all because they don’t want to be seen as wider.

Yes, there are visual competitive aspects to this, but surely if a slightly wider price comes with a reject rate that is a third of the competition, consumers would see that and trade accordingly? I thought we were heading in this direction a few years ago, and I know that some consumers do embrace the data and react accordingly, but clearly there are either too few noticing it, or the LPs are over-eager on last look. These latter two camps are where we should be focusing our attention, and the spearhead of those investigating should be the liquidity consumers. The data is available, how about using it?

colin@thefullfx.com